[lcoreyl]

This is my 4th year with a team that averages 7-8 "full time" members.
Build Season
Design
I'm guessing that small team = poor team = poor resources in most cases, so I definitely agree with the idea above of avoiding the best robot design so that you can end up with a robot that actually works. I guess the struggle is making sure that the kids who are making final design decisions fully understand how many teams each year can't achieve the design they come up with. If you haven't ever built something in your plans, get onto chiefdelphi and ask around if it's a bad idea to try it for the first time in build season.
CAD
We've never used it and every year I get tired of hearing myself harp on to the kids that they need to learn it (and offer to learn it with them). I told the non-seniors that if at least 1 of them doesn't complete an offseason Inventor project I won't be helping anymore (which likely means no team). I see CAD as addressing a few problems:
1) no/few mentors and kids who aren't mechanically savvy: now instead of "figure this out and build it", you can say "build this"
2) Kids who've missed a few days and don't know what's going on: they can more easily look and see what needs to get done. This year I've had kids who had no idea what our plan for shooting was 2 weeks after we'd basically decided.
3) Shooting too high with design: the realities of what needs to happen are easier to see if you've CADed things out detailed enough
Communication
If you can't spare the kids and mentors to be the project manager, you need to have something in place so a kid/mentor can see where you are, and what the current plan is. This year I created a forum online and made categories for strategy and all subsystems, then within them had our current design and any further questions that still needed to be researched/prototyped/etc. Admittedly it didn't work because I didn't have buy in from the kids (they rarely ever even looked there), but I think it has the potential to work. I might force its use...
Newbs
Unless you are in a really serious time crunch, try to make sure the newbs are the ones actually doing--the programmer at the keyboard, the person drilling, etc. Otherwise your experienced people all graduate, and now you have serious problems.
Off Season
-get kids to work on some project to increase their knowledge on one topic. Programming, pneumatics, CAD. An advanced kid(s) could also look at other robots they've seen at competitions or on chiefdelphi and try building parts of them--drivetrain, arms, lifts, etc. New programmers really should go through a lot of work with the cRIO before build season as well.
-Do as much as you can in the community to get some press, make sure any PR material always says that you are looking for mentors and/or financial support.
If you actually get some money
-8020 T-slot aluminum. Even our mechanically challenged ("what are vice grips?") can prototype with this stuff. We also have built our frames from it the last 2 years. Also, its reusable, so long term it's cost effective.
-Arduino sparkfun kits. These have been great because they introduce programming in a way that relates to what they will do on the robot, and to the motivated student needs little to no help from anyone else. I've even used these during build season to give a student something to do where otherwise they might leave because they don't feel like they can do anything on the robot.